RAIMS Innovation Fellow (OIP)
100-Day Research & Build Program — MSc/PhD Graduates
Tracks: EXTRACT
• CUSTOMIZE
• LOAD
Program: Orcawise Innovation Program (OIP)
Focus: Development of RAIMS (Responsible AI Management System)
About the Program
The Orcawise Innovation Program (OIP) is a 100-day research and build work experience program designed for MSc and PhD graduates looking to transition to a career in Responsible AI Management Systems (RAIMS)
As a Fellow, you'll join a small, high-impact cohort working on the development of RAIMS - the world's first operational framework for responsible AI governance, aligned with ISO/IEC 42001 and the EU AI Act. Your work will be applied directly to the internal tooling, standards, and systems behind and the broader Orcawise ecosystem.
Tracks You Can Join:
1. EXTRACT – Data Grounding for Trust
You'll build the trusted data layer for RAIMS:
* Source legally usable knowledge (laws, standards, docs)
* Ingest and process PDFs, HTML, CSVs
* Chunk, tag, and store content in a structured vector database
* Design gold-standard test sets and evidence-verified answers
* Evaluate grounding precision and citation fidelity
Outcome: A well-indexed, evaluable RAG pipeline that powers explainable answers in
2. CUSTOMIZE – Custom LLM, Prompt & Behavior Engineering
You'll design and test prompt behaviors under governance constraints:
* Create role-based system and user prompts with embedded metadata
* Apply advanced reasoning frameworks (ReAct, Chain-of-Thought, etc.)
* Implement safety refusals, guardrails, and content moderation logic
* Compare in-context learning vs. fine-tuning and example stores
* Optimize for accuracy, consistency, and governance traceability
Outcome: A catalog of prompts, templates, and control strategies aligned with RAIMS principles.
3. LOAD – Deploy & Govern
You'll build the agent interface and operations layer:
* Develop chat UI (persona selector, streaming, citations, feedback)
* Implement agent orchestration using functions, extensions, and RAG
* Log prompt metadata, cost, latency, and risk flags
* Build audit-export features and RBAC/SSO login
* Ensure traceability, fallback logic, and real-time observability
Outcome: A working prototype of with full governance tracking and compliance readiness.
What You'll Do (All Tracks)
* Scope and deliver a 100-day project with real-world impact
* Produce a demo + governance pack (logs, risks, tests)
* Track performance (accuracy, safety, latency, cost)
* Document decisions, architectures, and learning outcomes
* Share insights through a summary blog or lightning talk
Who We're Looking For
Required
* MSc or PhD (or near completion) in: AI, Data Science, Law+Tech, IS, HCI, Statistics, Engineering, or Policy
* Solid Python (data workflows, scripting, basic API usage)
* Strong writing and documentation skills
* Comfortable with experimentation, iteration, and self-direction
Nice to Have
* Experience with LLMs, prompt engineering, LangChain or similar
* Familiarity with RAG, embeddings, vector databases
* Interest in Responsible AI, ISO 42001, or the EU AI Act
* Cloud experience (GCP/AWS), front-end frameworks (for LOAD)
Program Milestones - How it Works
The OIP is running constantly with daily training sessions and work experience schedules. Join the community and log into the daily track sessions where you will be expected to hit the ground running for a demanding work experience.
What You'll Gain
* Certificate of Completion
* Verification of your contribution to the work experience program
* Experience designing responsible AI systems in production
* Deep understanding of RAIMS, ISO 42001, and regulatory-aligned prompt engineering
* Daily mentorship from Orcawise engineers, governance leads, and researchers
* A real-world project you can showcase as part of your portfolio or thesis
* Public credit (optional) for your contribution via blog/talk/demo
Job Type: Part-time
Expected hours: 20 per week
Application question(s):
* Can you commit a minimum of 20-hours per week to a 100-day work experience program focused on RAI?
Education:
* Master's (preferred)
Work Location: Remote